The OCR -> schema.org pipeline would be useful, I've been doing it manually prior to importing to Mealie.